REVIEW Open Access

Utilization of risk-based predictive stabilitywithin regulatory submissions; industry’sexperienceMegan McMahon1* , Helen Williams2, Elke Debie3, Mingkun Fu4, Robert Bujalski4, Fenghe Qiu5, Yan Wu6,Hanlin Li7, Jin Wang8, Cherokee Hoaglund-Hyzer9 and Donnie Pulliam10

Abstract

Risk-Based Predictive Stability (RBPS) tools, such as the Accelerated Stability Assessment Program (ASAP) and othermodels, are used routinely within pharmaceutical development to quickly assess stability characteristics, especiallyto understand mechanisms of degradation. These modeling tools provide stability insights within weeks that couldtake months or years to understand using long-term stability conditions only. Despite their usefulness, theknowledge gained through these tools are not as broadly used to support regulatory filing strategies. This paperaims to communicate how industry has used RBPS data to support regulatory submissions and discuss theregulatory feedback that was received.

Keywords: Risk-based predictive stability, RBPS, Clinical shelf-life, Predictive, Regulatory feedback

IntroductionIn 2015, the International Consortium for Innovationand Quality in Pharmaceutical Development (IQ)launched a working group to focus on the use of Risk-Based Predictive Stability (RBPS) tools to optimizepharmaceutical development. The working group hasapproximately 50 members from 18 companies acrossthe pharmaceutical industry. The working group con-ducted an industry survey to understand sponsor compan-ies’ experiences using RBPS tools (Williams et al. 2017).The survey indicated that RBPS tools were being utilizedin a variety of applications across the development con-tinuum. A key learning was that of all the surveyed com-panies utilizing RBPS tools, approximately 55% of them

were leveraging the data in a regulatory capacity. Of thosethat reported utilization of RBPS data in regulatory sub-missions, the level of experiences exceeded 100 submis-sions in total from clinical development throughregistration. This paper provides a more in-depth look atthese submissions and the regulatory feedback that wasreceived. The ideas shared in this document reflect infor-mation shared during the 2018 AAPS and IQ Joint ShortCourse entitled “Science and Risk-Based Stability Strat-egies: Applications of Predictive Tools.”The International Council for Harmonization (ICH)

first recommended the adoption of Guideline Q1A (Sta-bility Testing of New Drug Substances and Products) in2000 (ICH 2003). Since that time, significant advances inthe science of stability have occurred. Improved model-ing tools coupled with appropriate protocols now enableconfident stability predictions without long-term data(Waterman and Adami 2005; Waterman et al. 2007;Oliva et al. 2012; Wicks et al. 2018; Williams et al.2018). Although these tools do not replace the trad-itional long-term and accelerated studies that are de-scribed in ICH Q1A, they enhance drug development by

© The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you giveappropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate ifchanges were made. The images or other third party material in this article are included in the article's Creative Commonslicence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commonslicence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtainpermission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

* Correspondence: megan.mcmahon@pfizer.comThis paper was developed with the support of the International Consortiumfor Innovation and Quality in Pharmaceutical Development (IQ, www.iqconsortium.org). IQ is a not-for-profit organization of pharmaceutical andbiotechnology companies with a mission of advancing science andtechnology to augment the capability of member companies to developtransformational solutions that benefit patients, regulators and the broaderresearch and development community1Pfizer Inc., Eastern Point Road, Groton, CT 06340, USAFull list of author information is available at the end of the article

AAPS OpenMcMahon et al. AAPS Open (2020) 6:1 https://doi.org/10.1186/s41120-020-00034-7

providing an increased understanding of the attributesthat influence stability earlier in development. These ap-proaches are well aligned with the science and risk-based approaches detailed in ICH Q8-–Q11 (ICH 2009;ICH 2005; ICH 2008; ICH 2012) and have been termedRisk-Based Predictive Stability (RBPS) (Williams et al.2017). Companies are utilizing these RBPS tools to en-able the development of medicines (Freed et al. 2016).The survey indicated that the majority of the regu-

latory experience has been in the early clinical devel-opment phase (Williams et al. 2017). Based on this,the working group published a proposed regulatorytemplate to standardize how RBPS data can be re-ported in a regulatory filing (Stephens et al. 2018) tosupport an initial clinical shelf life. This paper isintended to expand on the regulatory templatepublication by sharing diverse case studies ofinstances where RBPS tools have been used to sup-port regulatory filings. The case studies illustrate howRBPS tools can be used throughout all stages ofdevelopment. The industry survey indicated that mostcompanies using RBPS tools are utilizing AcceleratedStability Assessment Program (ASAP) modeling (Wil-liams et al. 2017). The case studies reflect the pre-dominant use of ASAP; however, other predictivemodels could support similar regulatory strategies.The presented case studies are summarized in Table 1.

The majority focus on early clinical development, whichreflects how industry most commonly uses RBPS toolsin regulatory submissions. Where RBPS is used to pre-dict a retest date/shelf life, the predictions are alwayssupported by a 95% confidence interval. However, mostcompanies take a conservative approach to retest/shelf-life dating and assign a retest/shelf-life shorter than theprediction supports. Typically, if RBPS data alone is usedto establish the retest/shelf-life, a 1–2 year maximum isapplied.The case studies are presented in order of the phase of

development. Most of these case studies utilize RBPStools to predict degradation, but RBPS tools can also beused to predict dissolution. Case studies were providedby eight pharmaceutical companies. Additionally, thecollective regulatory experiences from 5 companies werepooled and analyzed with respect to global regulatoryacceptance of RBPS strategies.

Case studies – phase 1Case study 1: setting an initial retest period for drugsubstance and an initial shelf-life of the drug productAccelerated Stability Assessment Program (ASAP) stud-ies on drug substance and drug product were performedin 2014 to support a first-in-human (FIH) clinical trialwith an oral solution. The 14-day ASAP study for thedrug substance included seven temperature and

humidity conditions ranging from 40 °C/75%RH to80 °C/10% RH. The ASAP study for the drug product(oral solution) was one month long, and was performedwith five different temperatures ranging from 25 °C to70 °C. In both cases, assay and chemical degradationwere modeled. For the oral solution, the ASAP was per-formed in closed vials and for the drug substance, theASAP was performed in an open dish.The ASAP predictions and 1-month of stability data at

the proposed long-term storage condition on the drugsubstance were used to support an initial retest period of12 months, although the modeling predicted a retestperiod of at least 3 years. For the oral solution, a shelf-life of 6 months when stored at 2–8 °C was proposedbased on ASAP data only since no out-of-specificationtest results were predicted for at least 1 year at 5 °C andat least 6 months at 25 °C/60%RH storage conditions. Inthe IMPD, a commitment was made to perform trad-itional long-term and accelerated stability monitoring onboth drug substance and drug product and the stabilityprotocols were included. The submission was filed inBelgium in 2014 and was accepted without any relatedregulatory queries. The ASAP predictions were subse-quently confirmed by the available long-term stabilitydata.

Case study 2: leveraging RBPS to support shelf-life for anew strength of an existing tablet formulationIn 2015, a Phase 1 clinical trial was proposed requiring anew lower strength version of an existing tablet formula-tion. The existing tablets were known to be chemicallystable with a 5-year shelf-life based on five years of sta-bility data at 25 °C/60% RH and significant product un-derstanding and development knowledge. The new lowstrength tablet was formulated utilizing a common gran-ule with the existing tablets and therefore expected toalso be chemically stable. An ASAP study incorporatingfive temperature and humidity conditions from 50 °C/75% RH to 70 °C/75% RH for up to 4 weeks was per-formed using a crushed tablet, as the worst-case scenariodue to increased surface area. Chemical degradation wasmodeled based on one main degradant as the shelf-lifelimiting attribute (SLLA). Predictions did not considerpackaging (worst case).The ASAP predictions were used to support a 3-

year shelf-life claim, for the new low strength tablet,along with 3 months long-term stability data on thenew strength and previous stability data on the exist-ing higher strength formulations. The submission wasfiled in the USA, UK, France, Italy, Turkey, Egypt,Lebanon and Kenya in 2015 and was accepted with-out any related regulatory queries. The ASAP predic-tions were subsequently confirmed by the availablelong-term stability data.

McMahon et al. AAPS Open (2020) 6:1 Page 2 of 11

Case study 3: leveraging RBPS to support a change in thecapsule shellIn late 2016, a Phase 1 capsule formulation was reformu-lated to change from a gelatin capsule shell to a hy-droxypropyl methyl cellulose (HPMC) capsule shell. AnASAP study was performed to predict a shelf-life for thenew encapsulated formulation using a representativebatch. The 8-week ASAP study incorporated six

temperature and humidity conditions ranging from50 °C/30% RH to 70 °C/75% RH and chemical degrad-ation was modeled based on the formation of threedegradants, to identify the shelf-life limiting degradant.The ASAP predictions were submitted to the USA in

2017 with one month of long-term stability data for theHPMC capsules to support a 12-month shelf-life underambient conditions. This was accepted with no related

Table 1 Overview of case studies

Casestudyno.

Description Phase Material Submission Content inAddition to PredictiveDataa

Territories

Accepted Questioned

1 Setting initial retest period andshelf-life

Phase 1 Drugsubstanceand oralsolution

Limited long-term datafor drug substance

Belgium

2 Setting shelf-life of formulationvariant

Phase 1 Tablet Limited long-term data USA, UK, France, Italy, Turkey, Egypt,Lebanon, Kenya

3 Setting shelf-life of formulationvariant

Phase 1 Capsule Limited long-term data USA

4 Setting initial retest period andshelf-life

Phase 1 Drugsubstanceand IVsolution

None Netherlands, Germany (drugsubstance)

UK,Germany(drugproduct)

5 Setting initial retest period andshelf-life

Phase 1 Drugsubstanceand tablet

None USA UK

6 Setting initial shelf-life Phase 1 Tablet None Belgium, Moldova, Bulgaria, Georgia France,Spain

7 Setting initial shelf-life Phase 1 Powder fororalsolutionand tablet

None Germany

8 Setting shelf-life of formulationvariant

Phase 1 Capsule None USA, Spain, France

9 Setting initial shelf-life Phase 2 Tablets None USA, Belgium

10 Setting initial shelf-life Phase 2 Modifiedreleaseformulation

None USA, Belgium

11 Setting shelf-life of formulationvariant

Phase 2 Tablet None USA, Spain, France Germany

12 Justify storage condition for drugsubstance with a new process,and predict shelf-life for a newformulation

Phase 2 DrugSubstanceand Tablet

Limited long-term datafor drug substance, Lim-ited long-term data fordrug product

USA

13 Setting shelf-life of formulationvariant

Phase 2 Tablet Limited long-term data UK (but with reduced shelf-life)

14 Impact of drug product processchange on shelf-life

Phase 3 Tablet Limited long-term data USA, Canada, Russia, Taiwan, Poland,Ukraine, Japan

15 Setting shelf-life of formulationvariant

Phase 3 Capsule Supportive informationfrom previousformulations

Belgium, Brazil, Canada, China, India,Denmark. Finland, France, Ireland,Italy, Netherlands, Portugal, Spain,Taiwan, Turkey

UK, Korea,Germany

16 Justification of strategy for watertesting on stability

Registration Tablet Predictive dissolutionmodel

USA, EU

17 Justification of degradant end oflife specification

Registration Tablet Limited long-term data USA, EU, Canada, Australia, AsiaPacific and Latin America

aDoes not reflect additional data provided in response to regulatory queries

McMahon et al. AAPS Open (2020) 6:1 Page 3 of 11

regulatory queries. The ASAP predictions were subse-quently confirmed by the available long-term stabilitydata.

Case study 4: setting an initial retest period for drugsubstance and an initial shelf-life for a parenteral drugproductDuring 2017, ASAP studies were performed for the firstPhase 1 submission for a new drug substance and thedrug product formulation, an IV solution. The ASAPstudy performed for the drug substance utilized sixtemperature and humidity conditions ranging from50 °C/75% RH to 80 °C/30% RH. No degradation was ob-served during the 12-week drug substance ASAP study,so no modeling was performed. A similar range of tem-peratures were used for the drug product 6-week ASAPstudy, using a representative batch, and the data used topredict an initial shelf-life based on formation of themain degradant.The ASAP data for the drug substance and drug prod-

uct were presented in the application with predicteddegradation for the drug product. The ASAP data wereused to support a 12-month retest period for the drugsubstance under ambient conditions and a 12-month forthe drug product at 5 °C. Commitments to initiate trad-itional long-term and accelerated stability studies forboth the drug substance and drug product were made,but no long-term stability data were presented. The sub-mission was filed in the UK, Germany and theNetherlands. It was accepted with no related queries inthe Netherlands. The drug substance retest period wasaccepted by Germany, but the drug product shelf-lifeclaim was questioned and a 6-month shelf-life was sug-gested. In response, the 1 and 3-month long-term stabil-ity data for the drug product was presented to supportthe 12-month shelf-life claim. The MHRA requestedlong-term stability data for both drug substance anddrug product, which was subsequently provided. In theresponses to both Germany and the UK, the 1 and 3-month long-term stability data for the drug product wasoverlaid with the ASAP predictions at both 5 °C and25 °C to demonstrate the accuracy of the predictions andthe applications were approved. The ASAP predictionswere subsequently confirmed by the available long-termstability data.

Case study 5: setting an initial retest period for drugsubstance and an initial shelf-life for drug productASAP studies were performed during 2017 to supportthe initial Phase 1 submission for a new drug substanceand drug product tablet formulation, using representa-tive batches. The 6-week ASAP study incorporated sixtemperature and humidity conditions ranging from50 °C/75% RH to 80 °C/30% RH for the drug substance

and drug product. Minimal degradation was observed inthe drug substance ASAP study, so modeling was notpossible. Shelf-life predictions were performed for thedrug product based on the main degradant.The ASAP data for drug substance and drug product

were presented in the submission with predicted degrad-ation for the drug product. The ASAP data supported aninitial 12-month retest period for the drug substanceand an initial 12-month shelf-life for the drug productunder ambient conditions. Commitments to completetraditional long-term and accelerated stability studies forboth the drug substance and drug product were stated,but no long-term stability data were presented. The sub-mission was filed in the USA and UK. The US FDA hadno related queries, but UK MHRA requested 3 monthslong-term stability data for both the drug substance anddrug product. The requested data was supplied resultingin an agreement with the 12-month shelf-life and nodelay to the clinical study start. The ASAP predictionswere subsequently confirmed by the available long-termstability data.

Case study 6: setting an initial shelf-life for drug productIn support of a Phase 1 clinical trial, a 30-day ASAPstudy was performed applying 6 different temperatureand humidity conditions ranging from 40 °C/75% RH to70 °C/75%RH, on newly formulated tablets. During theASAP study, no degradation was observed, so modelingwas not possible. A shelf-life of 12 months was proposedfor the tablets. The submission was filed in 2015 toBelgium, Spain, France, Moldova, Bulgaria and Georgiaand was accepted in all countries, except for France andSpain where additional long-term data (3 months) wasrequested to approve the 12-month shelf-life. The ASAPpredictions were subsequently confirmed by the availablelong-term stability data.

Case study 7: setting an initial retest period for drugsubstance and an initial shelf-life for multiple drug productsA new Phase 1 clinical trial was proposed using a pow-der for oral solution (PfOS), consisting of drug substancein a bottle formulation and a tablet formulation. A 12-month shelf-life was required for both formulations toenable the clinical study. ASAP studies at fourtemperature and humidity conditions ranging from60 °C/50% RH to 70 °C/75%RH were conducted for upto three weeks on the drug substance and the tablet. Nodegradation was observed. The ASAP data were submit-ted to Germany in 2014 to request an initial 12-monthshelf-life for the PfOS and the tablet. The submissionalso included a commitment to monitor clinical batchesusing a traditional long-term and accelerated stabilityprotocol. No long-term data were provided in the initialsubmission.

McMahon et al. AAPS Open (2020) 6:1 Page 4 of 11

The Agency responded that “The submitted stabilitydata are inadequate for demonstrating a shelf-life of 12months for both investigational preparations. Additionalresults should be submitted which at least demonstratethe stability of the active ingredient and the tablets after3 months at 40°C/75%RH”. At the time the query wasreceived, the 3-month results on the tablet were avail-able and submitted together with 3-month drug sub-stance data at 40 °C/75%RH. No further queries werereceived from the Agency and the 12-month shelf-lifewas approved. The ASAP predictions were subsequentlyconfirmed by the available long-term stability data.

Case study 8: setting an initial shelf-life for a newformulation for an ongoing clinical studyIn 2013, a formulation was changed from an oral solu-tion to a capsule for an ongoing Phase 1 clinical trial. Atthe time of submission of the new formulation, no long-term stability data were available. A 14-day ASAP studyincorporating five temperature and humidity conditionsranging from 40 °C/75%RH to 70 °C/50% RH was per-formed. Assay and chemical degradation were modeled.Packaging predictions were also included.The ASAP predictions were used to support an initial

12-month shelf-life for the new capsule formulation.The submission included ASAP data with a commitmentto perform long-term stability on a representative cap-sule batch. The application was approved in the USA,France, and Spain in 2013. No regulatory queries werereceived on the proposed shelf-life. The ASAP predic-tions were subsequently confirmed by the available long-term stability data.

Case studies – phase 2Case study 9: setting an initial shelf-life for multiple tabletstrengths (not common blend)Clinical studies for a tablet formulation at threestrengths manufactured at varying drug loads wereplanned for the USA and Belgium in 2016. The submis-sions did not include long-term stability data but didprovide a commitment to place representative clinicalbatches of each strength on long-term stability. AnASAP study was performed on a development batch ofcompacts for the strength with the highest excipient todrug ratio, maximizing the potential for drug substancedegradation.The ASAP experiment was run for 14 days in open

containers incorporating six temperature and humidityconditions ranging from 50 °C/75% RH to 80 °C/40%RH. The results included levels for two primary degra-dants. The data was modeled and predicted that roomtemperature storage may not provide a suitable shelf-life, but a shelf-life of 12 months could confidently be setfor all three strengths when stored at 5 °C. These data

and predictions were submitted to both health author-ities. No queries were received. Since the initial submis-sions, long-term data has supported the storage of thesupplies under refrigeration and room temperature.

Case study 10: setting an initial shelf-life for a modifiedrelease drug productA modified-release (MR) formulation at two strengths(the strengths were composed of different levels of drugsubstance to produce a common tablet weight) was in-troduced into clinical studies in the USA and Belgium.At the time of this 2015 submission, no long-term datawas available, and the initial shelf-life was supported byan ASAP study.The ASAP study was run on both tablet strengths for

60 days using eight temperature and humidity condi-tions, ranging from 30 °C/5%RH to 60 °C/5%RH and50 °C/52%RH. A single degradant was identified as theSLLA and was modeled to predict that the MR tabletswould be stable for at least 12 months when stored at25 °C. An initial shelf-life of 12 months was proposed.The results of the ASAP study were submitted alongwith a commitment to place a representative clinicalbatch of each strength on long-term and accelerated sta-bility. No related regulatory queries were received. TheASAP predictions were subsequently confirmed by theavailable long-term stability data.

Case study 11: setting an initial shelf-life for a newformulation introduced into an ongoing clinical studyIn 2017, clinical application amendments were filed tointroduce an additional tablet formulation. Two tabletstrengths with a different formulation were already ap-proved for clinical use. A different strength was pro-posed for the new formulation and no long-term datawas available for the clinical amendments.A 4-week ASAP study was conducted to support the

initial clinical shelf-life for the new formulation and re-sults were provided in the regulatory submission. Thestudy was conducted on a development strength tablet(0.25 mg instead of the proposed clinical strength of 15mg). The development strength (0.25 mg) contained ahigher excipient to drug ratio than the proposed clinicalstrength, thereby providing a worst-case assessment fordegradation. They were stored in open glass bottles andexposed to eleven separate temperatures, humidities anddurations. The conditions varied from 50 °C/75% RH to80 °C/40% RH and 70 °C/75% RH.The stability samples were subsequently evaluated for

appearance and concentrations of the main degradant.The main degradant was used to predict the shelf-lifeand predictions supported an initial shelf-life of at least12 months.

McMahon et al. AAPS Open (2020) 6:1 Page 5 of 11

This application was filed to the USA, Spain, Germanyand France. For Spain, the USA and France, an initialshelf-life of 12 months was accepted without relatedregulatory queries. Germany queried the assignment of a12-month shelf-life without long-term data. In response,3 months of long-term data were provided at the pro-posed storage conditions and accelerated conditions(40 °C/75% RH) and the shelf-life was approved. TheASAP predictions align with the available long-term sta-bility data.

Case study 12: justification of storage conditions for drugsubstance and setting an initial shelf-life for drug productASAP studies on drug substance and drug product wereperformed to support a Phase 2 clinical program in2018. The drug substance was manufactured via a newprocess and a new drug product was introduced. In bothcases, the studies were performed on representative de-velopment batches in an open dish configuration, andchemical degradation kinetics were modeled on themajor degradant. The 42-day ASAP study for the drugsubstance included 5 temperature and humidity condi-tions ranging from 50 °C/80%RH to 80 °C/0%RH. Fordrug product, the 100 day ASAP study was performedbased on 5 different temperature and humidity condi-tions ranging from 50 °C/60%RH to 70 °C/10%RH. Thelong study duration was due to the need to use lessstressful conditions because of potential form changes athigher temperatures. The ASAP predictions and 9-month long-term stability on clinical batches of drugsubstance were used to justify refrigerated storage condi-tion. For drug product, the ASAP model was used tosupport a predicted 2-year shelf-life. The ASAP resultswere submitted to FDA, together with 6 months of sta-bility on developmental batches and a commitment toperform long-term stability on the clinical batches. Thisrisk-based stability approach was accepted by the agencywithout stability data on clinical batches. Long-term sta-bility studies on the clinical batches are ongoing. TheASAP predictions align with the available long-term sta-bility data.

Case study 13: setting an initial shelf-life of a drug productIn 2018 a new drug product formulation was plannedfor use in a Phase 2 relative bioavailability (rBA) study inthe UK. Subsequently, the same formulation would beused in Phase 3 studies. A 28-day ASAP open-dish studywas conducted on a representative lot. Experiments in-cluded 9 temperature and humidity conditions rangingfrom 30 °C to 80 °C and 11%RH to 82% RH. Chemicaldegradation was modeled based on a major degradant.The ASAP model predicted acceptable stability in theclinical package for at least 1.6 years at ambient storageconditions.

The ASAP results were submitted to support a 12-month shelf-life for the new formulation. One month oflong-term stability data on a representative batch wassubmitted with a commitment to continue the confirma-tory long-term stability study. A commitment was alsomade to run a confirmatory long-term stability study ona clinical batch through the proposed shelf-life. The UKreviewer did not query the approach but reduced theshelf-life to 6months as part of the IMPD approval let-ter. There was no opportunity to provide additional sta-bility data during the review. The ASAP predictionsalign with the available long-term stability data.

Case studies – phase 3Case study 14: re-assessment of shelf-life after a drugproduct process changeDuring the Phase 3 development of a tablet formulationin 2016, a process change from wet granulation to rollercompaction was required. The product was known to bechemically stable, with two years long- term stabilitydata on the wet granulation product. An 8-week ASAPstudy was performed on a representative batch of theroller compacted formulation, utilizing six temperatureand humidity conditions ranging from 50 °C/75% RH to80 °C/30% RH. No degradation was observed in theASAP study, demonstrating the chemical stability of theproduct. A reduction in the rate of dissolution was ob-served in tablets exposed to high humidity. This trendwas consistent with previous studies of the wet granula-tion formulation; therefore, the tablets were stored withdesiccant. The results from both studies were used todemonstrate equivalence between the two formulationswith respect to stability attributes.The ASAP and dissolution data for the roller compac-

tion formulation was presented in the regulatory submis-sion to support a 2-year shelf-life alongside one monthof long-term stability data and 2-year stability data onthe wet granulation formulation. The submission wasfiled in the US, Canada, Russia, Taiwan and Poland andaccepted with no related regulatory queries. Long-termdata subsequently confirmed that the assigned shelf-lifewas appropriate.During 2017 a lower strength tablet of the same roller

compaction formulation (common blend) was requiredfor clinical studies. The ASAP and dissolution studieswere repeated for the low strength formulation and simi-lar results obtained. The data was used to demonstrateequivalence with the higher strength formulation andthe wet granulation formulation. The ASAP and dissol-ution data were presented in the regulatory submissionin the absence of long-term data on the low strengthformulation, but alongside 6 months stability data on thehigher strength formulation. A commitment was madeto set down long-term stability studies for the low

McMahon et al. AAPS Open (2020) 6:1 Page 6 of 11

strength formulation in the submission. It was filed inUkraine, Russia, Canada, US, Taiwan, Poland and Japanand was accepted without related regulatory queries.The ASAP predictions were subsequently confirmed bythe available long-term stability data.

Case study 15: setting an initial shelf-life for a new drugproduct formulationIn 2010, IND and IMPD applications were filed globally(Belgium, Brazil, Canada, China, Denmark, Finland,France, Germany, India, Ireland, Italy, Korea,Netherlands, Portugal, Spain, Taiwan, Turkey, Ukraine,United Kingdom) to support a Phase 3 study using anew drug product formulation without traditional long-term and accelerated data on the new formulation. Thenew formulation was the intended commercial formula-tion and was a formulated immediate release (IR) gelatincapsule. Phase 1 and 2 formulations included powder(drug substance) in a capsule (PIC) and IR tablet. Stabil-ity data for the Phase 1 and 2 formulations were sup-portive, since the data provided long-term datasupporting chemical compatibility between the drugsubstance and most of the excipients being used in thecapsule formulation.A statistically designed 14-day ASAP study was per-

formed on a development lot of the capsules that wascompositionally equivalent to the proposed clinical cap-sules. The development lot was stored in open glass bot-tles and exposed to various temperatures, humiditiesand durations. Ranging from 50 °C/75%RH to 80 °C/75%RH and 5 to 14 days. Testing was performed on de-velopment batches containing the highest excipient todrug ratio for this product, maximizing possible drug-excipient interactions.The stability samples were evaluated for appearance,

levels of the main degradant, and total degradants. Allappearance data remained within specification with nosignificant trends observed. The main degradant was de-termined to be the shelf life limiting degradant at roomtemperature conditions. The degradant was modeled,and the results were used to predict an initial clinicalshelf life for the formulation. The model predicted anacceptable shelf life of at least 2 years, when stored at30 °C/75% RH.The predictive results along with supportive data from

the previous clinical formulations were filed to the coun-tries listed above, along with a commitment to evaluatethe stability of the clinical supplies through the durationof the clinical studies under long-term and acceleratedconditions. The initial shelf-life was assigned as 12months. This approach was accepted without queries inall countries except UK, Korea and Germany. Threemonths of long-term data were provided in response toqueries in these countries and the shelf-life was

approved. The ASAP predictions were subsequently con-firmed by the available long-term stability data.

Case studies: registrationCase study 16: justification of drug product strategy forwater testing on stabilityFor a recent immediate release formulation made up ofan active-containing amorphous solid dispersion, it wasdetermined that water content is not a critical quality at-tribute (CQA) and does not require analytical control.To justify this, in addition to a forced crystallizationstudy and open dish chemical stability studies, dissol-ution was evaluated for the impact of temperature andwater content. An ASAP study on tablet dissolution wasdesigned using seven accelerated temperatures and hu-midity conditions ranging from 50 °C/60%RH to 60 °C/75%RH for up to 2 weeks. Considerable changes in dis-solution were observed under highly stressed conditions.The ASAP data along with data from the 40 °C/75%RHopen dish study at 3 and 6months were used in the datamodeling. With the fitted modified Arrhenius equation,along with the water sorption isotherms of the tabletsand packaging configurations, dissolution profiles werepredicted throughout the product shelf-life under anygiven storage condition and initial tablet water content.The model was then verified with available real time sta-bility data and agreed with the observed real time data(Li et al. 2016).The ASAP study and dissolution prediction model

were presented in Section 3.2.P.2.3 ManufacturingProcess Development of the product marketing applica-tion submitted to the USA and EU. The results demon-strated that under the proposed packaging and storageconditions the water content of the tablets would notimpact dissolution justifying water testing is not requiredon stability. The data and strategy were accepted by theagencies with no related review queries. The ASAP pre-dictions were subsequently confirmed by the availablelong-term stability data.

Case study 17: justification of degradant end of shelf-lifespecificationA Product Characterization Study (PCS) (Lavrich 2016;Wu et al. 2018) is a systematic study that explores theproduct stability as the product is exposed unprotectedto various constant environmental conditions to under-stand the impact of storage conditions and formulationcomposition / manufacture process on the stability ofproduct over the anticipated shelf life. It is used to buildpredictive stability models and to inform the designspace, control strategy, specification setting, package se-lection, shelf-life and post-approval changes.A PCS was performed to assess the impact of moisture

and temperature on the degradation of an extended

McMahon et al. AAPS Open (2020) 6:1 Page 7 of 11

release tablet formulation over an anticipated 2-yearshelf- life. The study was conducted under open dishconditions by exposing drug product stored at 25 °C,30 °C and 40 °C over a range of humidity conditions(15%, 35%, 45%, 60% and 75%RH) with sampling at vari-ous time points over 2 years. PCS data revealed that theAPI in drug product was sensitive to moisture with hy-drolysis degradation the primary degradation pathway.An example of relevant hydrolysis degradation results ofthe PCS study at 25 °C is provided in Fig. 1.The data revealed that levels of the hydrolysis product

were suitably controlled in samples stored below35%RH. The selected high-density polyethylene (HDPE)bottle packages were designed to maintain a headspaceRH of less than or equal to 35% for one potency and30% for two other potencies over the course of 2 yearswhen stored at room temperature conditions across allfour climatic zones. The headspace moisture is effect-ively controlled in bottle packages by adjusting the desic-cant loading to account for long-term moisturetransmission across the primary package barrier, as wellas the moisture introduced via the product mass.To support the proposed shelf-life degradation limit,

the PCS predictions were filed in NDA drug productsections 3.2.P.5.6 Justification of Specification(s) and3.2.P.8.1 Stability Summary and Conclusion in multiplemarkets including the US, EU, Canada, Australia, AsiaPacific, and Latin America starting in 2011. The dataand strategy were accepted by the agencies with no re-lated regulatory queries.

Discussion of regulatory feedback and regulatorytrendsExperience using risk-based predictive tools to set aninitial clinical shelf-life reaches back to 2009. Overall,the regulatory acceptance of this strategy has been high.The collective experience of five companies involvedwith authoring this publication is summarized in Fig. 2.Here, the bars on the left (blue) represent the number ofclinical applications that were submitted and approvedutilizing only predictive stability to support the initialshelf-life. Within these submissions, commitments werealso made to conduct long-term and traditional acceler-ated stability, but data from those studies were not avail-able for the initial submission. The middle bars (red)represent the portion of those clinical submissions wherethe regulatory reviewer required submission of data fromthe traditional stability study prior to granting the re-quested shelf-life. The bars on the right (green) repre-sent those submissions where the proposed shelf-life wasnot accepted. In the cases where the shelf-life was notaccepted, the reasons provided link the non-acceptanceto lack of long-term stability data, rather than relatingthe non-acceptance to specific concerns with the modelused. Over the years, most of the regulatory applicationsthat submitted only predictive data to support the initialshelf-life were accepted. As illustrated by the red bars in2016 and 2017, the number of requests for additionaldata to substantiate the initial shelf-life appears to be in-creasing. And, although the contributing companies havenot experienced the rejection of their application based

Fig. 1 Predicted Hydrolysis as a Function of Humidity at 25 °C. The Regression Line and the 95% Confidence Interval are Presented. The ProposedSpecification Limit for the Hydrolysis Product is 0.30%

McMahon et al. AAPS Open (2020) 6:1 Page 8 of 11

on this strategy, there are a small number of occurrenceswhere the proposed shelf-life was not accepted.There has been an apparent decrease in regulatory ac-

ceptance of the use of RBPS to set initial shelf-life since2016. A higher percentage of submissions were queriedfor the provision of data from traditional studies. In2018, some companies were not granted the requestedshelf-life after the provision of long-term data or werenot permitted the opportunity to provide additional data.

The recent trend may be in part due to the publicationof the 2017 EMA clinical guideline (European medicinesagency 2017) which provides increased instruction onhow companies should set and extend shelf-life in theclinical stage and does not acknowledge RBPS as a toolto set an initial shelf-life. When using risk-based strat-egies to set or extend a clinical shelf-life that do notalign with the recommendations in the guideline, com-panies should clearly articulate their justification.

Fig. 2 Regulatory Experience Using RBPS to Set an Initial Shelf-life

Fig. 3 RBPS Regulatory Experience by Country (2009–2018). *USA Bar is truncated (35 accepted without requests for additional data)

McMahon et al. AAPS Open (2020) 6:1 Page 9 of 11

The data since 2009 was also analyzed by country andthese results are presented in Fig. 3. The bottom (red)bar represents the number of submissions made to agiven country with only RBPS data for the proposedclinical material(s). The top (blue) bar represents thenumber of submissions that were approved without theprovision of additional traditional stability data. The ana-lysis indicates that RBPS approaches for setting initialshelf-life have generally been accepted by most coun-tries. However, several countries do not embrace this ap-proach. This could point to a need for furthertransparency and discussion within submissions on thedetails of the model and the assumptions that weremade. Harmonization amongst industry on how thesedata are presented and utilized may also enable broaderregulatory acceptance. A proposed approach for this wasrecently published (Stephens et al. 2018).Beyond the clinical space, the case studies presented

where RPBS data were used to support registration ap-plications illustrate the broad range these tools have forimpacting development knowledge. These case studiesdemonstrate how impactful RBPS data can be as sup-portive development data. Based on feedback received inthe Industry Survey (Williams et al. 2017), companiesare routinely using RBPS to identify optimal formula-tions, for packaging selection and to demonstrateequivalence to support changes. Discussion of theseRBPS data within a marketing application would furtherunderwrite fundamental product understanding.

ConclusionThe case studies presented in this paper demonstratethat RBPS can be and is being used to set an initial retestperiod for clinical drug substance, an initial shelf-life formany clinical formulations, and to support changes (e.g.,to demonstrate equivalence after a process/formulationchange or to assess the risk to stability after a packagingchange) (Timpano et al. 2018). Where available, long-term data for each case are consistent with the RBPSpredictions. The use of predictive tools to set initial re-test/shelf-life is well-precedented and has been used rou-tinely by some companies for almost ten years. Itenables shelf-life to be set in a more expedient manner.This, in turn, enables submission of clinical applicationsin a timelier fashion. The regulatory risks to this ap-proach are presented above and can be mitigated by en-suring the appropriate duration of long-term data isavailable for query responses to further substantiate theproposed shelf-life.Analysis of the collective recent regulatory experience

(refer to Fig. 2) suggests an apparent decrease in regula-tory acceptance of the use of RBPS to set initial shelf-lifesince 2016. Companies with experience using these toolsfeel this is at odds with RBPS being a powerful and

science-based tool for gaining a more rapid and thor-ough understanding of the stability characteristics of adrug substance and drug product. The depth of stabilityknowledge gained through an RBPS study typically wellexceeds the information learned after, for example, thefirst 3-month time point of a long-term study where, formost well-designed products, changes are insignificant.Although the approach is not intended to replace long-term studies for setting shelf-life, it certainly comple-ments traditional long-term and accelerated stabilitystudies in that the modeling is ultimately validated bythe long-term data.The varied case studies presented herein demonstrate

that the use of RBPS tools can play an important role inregulatory submissions. In the clinical space, RBPS canlead to more rapid development timelines which pro-vides benefit for patients in need of new medications.The tools provide a key benefit to development and fun-damental product understanding. There is a clear oppor-tunity to continue to share this data and utilize it toconvey enhanced product understanding in regulatorysubmissions. Beyond the case studies presented here, thepotential regulatory applications for RBPS are many and di-verse. Moving forward, these tools could be used to bringmedicines to the market sooner, particularly when consider-ing long-term (12month) data per ICH can be the rate limit-ing step for a marketing application (European 2017). RBPStools could be used effectively post-approval to understandand convey risk to stability after a change to a process orproduct, rather than relying on collection of long-term data(Malcom 2018). In these cases, RBPS once again has the po-tential to provide value adding and incisive information in ashort amount of time. When used effectively, this datashould enable industry to make the best decisions through-out development and lifecycle management and enhanceregulatory reviewers’ confidence in industry’s abilities tounderstand the stability of our products and further protect-ing the patient.

AbbreviationsASAP: Accelerated Stability Assessment Program; HPMC: hydroxypropylmethyl cellulose; FIH: First in Human; ICH: International Council forHarmonization; IQ: International Consortium for Innovation and Quality;PCS: Product Characterization Study; PfOS: Powder for oral solution;RBPS: Risk-Based Predictive Stability

AcknowledgementsDennis Stephens, Abbvie.

Authors’ contributionsAll authors contributed case studies, ideas and information to assessregulatory feedback for this manuscript. HW authored the introduction tothe manuscript. MM authored the discussion and conclusions. All authorsreviewed and approved the manuscript.

FundingNot applicable; authors contributed case studies based on existing companyknowledge and experience.

McMahon et al. AAPS Open (2020) 6:1 Page 10 of 11

Availability of data and materialsNot applicable.

Competing interestsThe authors declare that they have no competing interests.

Author details1Pfizer Inc., Eastern Point Road, Groton, CT 06340, USA. 2AstraZeneca, SilkRoad Business Park, Macclesfield, Cheshire SK10 2NA, UK. 3JanssenPharmaceutica R&D, a Division of Janssen Pharmaceutica NV, Turnhoutseweg30, 2340 Beerse, Belgium. 4Sunovion Pharmaceuticals, 84 Waterford Dr,Marlborough, MA 01752, USA. 5Boehringer Ingelheim, 900 Ridgebury Road,Ridgefield, CT 06877, USA. 6Merck & Co., Inc., Kenilworth, NJ 08550, USA.7Vertex Pharmaceuticals, 50 Northern Avenue, Boston, MA 02210, USA.8Genentech, 1 DNA Way, South San Francisco, CA 94080, USA. 9Eli Lilly &Company, Lilly Corporate Center, Indianapolis, IN 46285, USA. 10Biogen, 5000Davis Dr, Durham, NC 27709, USA.

Received: 1 October 2019 Accepted: 17 April 2020

ReferencesEuropean federation of pharmaceutical industries and association (2017) EFPIA-

EBE White Paper on Expedited CMC Development: Accelerated Access forMedicines of Unmet Medical Need – CMC Challenges and Opportunities(Final Version - December 2017). https://www.efpia.eu/media/288657/efpia-ebe-white-paper-expedited-cmc-development-accelerated-access-for-medicines-of-unmet-medical-need-december-2017.pdf

European medicines agency (2017) EMA/CHMP/QWP/545525/2017 Guideline onthe requirements for the chemical and pharmaceutical qualitydocumentation concerning investigational medicinal products in clinicaltrials. https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-requirements-chemical-pharmaceutical-quality-documentation-concerning-investigational_en.pdf

Freed AL, Colgan ST, Kochling JD, Alasandro MS (2016) AAPS workshop:accelerating pharmaceutical development through predictive stabilityapproaches, April 4–5, 2016. AAPS Open 3:8

ICH: Guideline Q10 (2008) Pharmaceutical quality system, Step 4 version. https://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q10/Step4/Q10_Guideline.pdf

ICH: Guideline Q11 (2012) Development and manufacture of drug substances(chemical entities and biotechnological/biological entities), step 4 version.https://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q11/Q11_Step_4.pdf

ICH: Guideline Q1A(R2) (2003) Stability testing of new drug substances andproducts, Step 4 version. https://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q1A_R2/Step4/Q1A_R2_Guideline.pdf

ICH: Guideline Q8(R2) (2009) Pharmaceutical development, step 4 version. https://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q8_R1/Step4/Q8_R2_Guideline.pdf

ICH: Guideline Q9 (2005) Quality risk management, step 4 version. https://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q9/Step4/Q9_Guideline.pdf

Lavrich, D (2016) AAPS webinar: rapid development of robust stability modelsusing semi-empirical design space, 24 March 24, 2016 https://aapsopen.springeropen.com/articles/10.1186/s41120-017-0018-5#article-info

Li H, Nadig D, Kuzmission A, Riley CM (2016) Prediction of the changes in drugdissolution from an immediate-release tablet containing two activepharmaceutical ingredients using an accelerated stability assessmentprogram (ASAP prime®). AAPS Open 2:7

Malcom L (2018) Chapter 12: accelerated stability assessment Program (ASAP)applications in a postapproval environment. Accelerated predictive stability1st edition pp:231–241. doi:https://doi.org/10.1016/C2014-0-02298-8

Oliva A, Faria JB, Llabrés M (2012) An improved methodology for data analysis inaccelerated stability studies of peptide drugs: practical considerations.Talanta 94:158–166

Stephens D, Williams H, McMahon M, Qiu F, Hyzer CH, Debie E, Wu Y, Li H, WangJ (2018) Risk-based predictive stability for pharmaceutical development – aproposed regulatory template. Pharm Tech 42(8):42–47

Timpano RJ, Freed AL, Clement E, Masse K, Ryan K (2018) Chapter 9: acceleratedpredictive stability regulatory strategies. Accelerated predictive stability 1stedition pp:231–241. doi:https://doi.org/10.1016/C2014-0-02298-8

Waterman KC, Adami R (2005) Accelerated aging: prediction of chemical stabilityof pharmaceuticals. Int J Pharm 293(1–2):101–125

Waterman KC, Carella AJ, Gumkowski MJ, Lukulay P, MacDonald BC, Roy MC,Shamblin SL (2007) Improved protocol and data analysis for acceleratedshelf-life estimation of solid dosage forms. Pharm Res 24(4):780–790

Wicks BS, Lewis T, Khawam A (2018) Chapter 15 applications of ASAP to genericdrugs. Accelerated predictive stability 1st edition pp: 341–352. doi:https://doi.org/10.1016/C2014-0-02298-8

Williams H, Bright J, Roddy E, Poulton A, Cosgrove SD, Turner F, Harrison P,Brookes A, MacDougall E, Abbott A, Gordon C (2018) A comparison of drugsubstance predicted chemical stability with ICH compliant stability studies.Drug Develop Ind Pharm 45(3):379–386

Williams H, Stephens D, McMahon M, Debie E, Qiu F, Hyzer CH, Sechler L, Orr R,Webb D, Wu Y, Hahn D (2017) Risk-based predictive stability – an industryperspective. Pharm Tech 41(3):52–57

Wu Y, McMahon M, Regler B (2018) Applications of accelerated stability modelsin product development. PharmSci360 short course. https://www.conferenceharvester.com/Uploads/Documents/HarvesterLink-5946-748(41).pdf

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

McMahon et al. AAPS Open (2020) 6:1 Page 11 of 11